dtest1.cpp File Reference

generates data for precision test More...

#include "../homog2d.hpp"
#include <limits>
#include <iostream>

Include dependency graph for dtest1.cpp:

Macros
#define	NUMTYPE   long double

Functions
NUMTYPE	getRandom ()
int	main (int argc, const char **argv)

Variables
double	gk = 1000.
double	grmax = std::numeric_limits<double>::min()
double	grmin = std::numeric_limits<double>::max()

Detailed Description

generates data for precision test

This program will run 1 million of the following experiment:

• generate two random points in the range given by the first argument
• compute the line joining these two points
• compute the distance between the points and the line

Eventually, it will print several computed value:

1. range
2. min coordinate value
3. max coordinate value
4. mean distance value
5. max distance value
6. mean distance value (computed with the "running std" method)
7. standard deviation of distance value

Standard deviation is computed using a "single pass" algorithm, see https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance

To run that experiment on several ranges use the provided target:

$ make dtest1

Macro Definition Documentation

NUMTYPE

#define NUMTYPE   long double

Function Documentation

getRandom()

NUMTYPE getRandom

(

)

{
    double r = (2.0*rand()/RAND_MAX -1.) * gk;
    grmin = std::min( std::abs(r), grmin );
    grmax = std::max( std::abs(r), grmax );
    return r;
}

Here is the caller graph for this function:

main()

int main	(	int	argc,
		const char **	argv
	)

{
    auto nb = 1E6; // nb of runs
    char sep = ';';
    if( argc>1)
        gk = std::atof( argv[1] );
//  std::cout << "Running with k=" << gk << "\n";

    std::srand( time(0) );
    long double sum  = 0.;
    long double vmax = 0.;
    long double M2   = 0.;
    long double mean = 0.;
    for( auto i=0; i<nb; i++ )
    {
        Point2d_<NUMTYPE> pt1( getRandom(), getRandom() );
        Point2d_<NUMTYPE> pt2( getRandom(), getRandom() );
        auto li = pt1 * pt2;

        long double d = li.distTo(pt1);
        vmax = std::max( vmax, d );
        sum += d;

        auto delta = d - mean;
        mean += delta / (i+1);
        auto delta2 = d - mean;
        M2 += delta * delta2;

    }
    auto sigma = M2 / (nb-1);
    std::cout << std::scientific
        << gk << sep
        << grmin << sep
        << grmax << sep
        << sep << sum / nb
        << sep << vmax
        << sep << mean
        << sep << sigma
        << "\n";
}

Here is the call graph for this function:

Variable Documentation

gk

double gk = 1000.

grmax

double grmax = std::numeric_limits<double>::min()

grmin

double grmin = std::numeric_limits<double>::max()